The decision to accept or reject a VaR model based on backtesting results at the two-tailed 95% confidence level is less reliable with a 99% VaR model than with a 95% VaR model.

The 95% VaR model is less likely to be incorrectly rejected using backtesting than the 99% VaR model.

Using a 99% VaR model will lower the probability of committing both type 1 and type 2 errors.